Magnetic random access memory (MRAM) devices are showing increasing promise for utilization as memory storage devices of the future. MRAM is a type of digital memory in which digital bits of information comprise alternative states of magnetization of magnetic materials in memory cells. The magnetic materials can be thin ferromagnetic films. Information can be stored and retrieved from the memory devices by inductive sensing to determine a magnetization state of the devices, or by magnetoresistive sensing of the magnetization states of the devices. It is noted that the term “magnetoresistive device” can be utilized to characterize a memory device and not the access device, and accordingly a magnetoresistive device can be accessed by, for example, either inductive sensing or magnetoresistive sensing methodologies.
A significant amount of research is currently being invested in magnetic digital memories, such as, for example, MRAM's, because such memories are seen to have significant potential advantages relative to the dynamic random access memory (DRAM) components and static random access memory (SRAM) components that are presently in widespread use. For instance, a problem with DRAM is that it relies on electric charge storage within capacitors. Such capacitors leak electric charge, and must be refreshed at approximately 64-128 millisecond intervals. The constant refreshing of DRAM devices can drain energy from batteries utilized to power the devices, and can lead to problems with lost data since information stored in the DRAM devices is lost when power to the devices is shut down.
SRAM devices can avoid some of the problems associated with DRAM devices, in that SRAM devices do not require constant refreshing. Further, SRAM devices are typically faster than DRAM devices. However, SRAM devices take up more semiconductor real estate than do DRAM devices. As continuing efforts are made to increase the density of memory devices, semiconductor real estate becomes increasingly valuable. Accordingly, SRAM technologies are difficult to incorporate as standard memory devices in memory arrays.
MRAM devices have the potential to alleviate the problems associated with DRAM devices and SRAM devices. Specifically, MRAM devices do not require constant refreshing, but instead store data in stable magnetic states. Further, the data stored in MRAM devices will remain within the devices even if power to the devices is shutdown or lost. Additionally, MRAM devices can potentially be formed to utilize less than or equal to the amount of semiconductor real estate associated with DRAM devices, and can accordingly potentially be more economical to incorporate into large memory arrays than are SRAM devices.
Although MRAM devices have potential to be utilized as digital memory devices, they are currently not widely utilized. Several problems associated with MRAM technologies remain to be addressed. It would be desirable to develop improved methods for operation of MRAM devices.
FIG. 1 illustrates a fragment of an exemplary prior art construction 10 comprising an MRAM device 12. More specifically, construction 10 comprises a substrate 14 having a conductive line 16 formed thereover, and device 12 is formed over the conductive line.
Substrate 14 can comprise an insulative material, such as, for example, borophosphosilicate glass (BPSG), silicon dioxide and/or silicon nitride. Such insulative material can be formed over a semiconductive material, such as, for example, monocrystalline silicon. Further, various integrated circuit devices can be supported by the semiconductive material. In the construction of FIG. 1, substrate 14 is illustrated generically as a homogeneous mass, but it is to be understood from the discussion above that substrate 14 can comprise numerous materials and layers. In the event that substrate 14 comprises a semiconductive material, such semiconductive material can be, for example, monocrystalline silicon lightly-doped with a background p-type dopant. To aid in interpretation of the claims that follow, the terms “semiconductive substrate” and “semiconductor substrate” are defined to mean any construction comprising semiconductive material, including, but not limited to, bulk semiconductive materials such as a semiconductive wafer (either alone or in assemblies comprising other materials thereon), and semiconductive material layers (either alone or in assemblies comprising other materials). The term “substrate” refers to any supporting structure, including, but not limited to, the semiconductive substrates described above.
Conductive line 16 can comprise, for example, various metals and metal alloys, such as, for example, copper and/or aluminum.
The MRAM device 12 formed over line 16 comprises three primary layers, 18, 20 and 22. Layers 18 and 22 comprise soft magnetic materials, such as, for example, materials comprising one or more of nickel, iron, cobalt, iridium, manganese, platinum and ruthenium. Layer 20 comprises a non-magnetic material. The non-magnetic material can be an electrically conductive material (such as copper) in applications in which the MRAM is to be a giant magnetoresistive (GMR) device, or can be an electrically insulative material (such as, for example, aluminum oxide (Al2O3) or silicon dioxide), in applications in which the MRAM device is to be a tunnel magnetoresistive (TMR) device.
Layers 18 and 22 have magnetic moments associated therewith. The magnetic moment in layer 18 is illustrated by arrows 19, and the magnetic moment in layer 22 is illustrated by arrows 21. In the shown construction, the magnetic moment in layer 22 is anti-parallel to the magnetic moment in layer 18. Such is one of two stable orientations for the magnetic moment of layer 22 relative to that of 18, with the other stable orientation being a parallel orientation of the magnetic moment in layer 22 relative to the moment in layer 18. One of layers 18 and 22 can have a pinned orientation of the magnetic moment therein, and such can be accomplished by providing a hard magnetic layer, or in other words a permanent magnet (not shown) adjacent the layer. The layer having the pinned magnetic moment can be referred to as a reference layer.
In operation, MRAM device 12 can store information as a relative orientation of the magnetic moment in layer 22 to that in layer 18. Specifically, either the anti-parallel or parallel orientation of the magnetic moments of layers 18 and 22 can be designated as a 0, and the other of the anti-parallel and parallel orientations can be designated as a 1. Accordingly, a data bit can be stored within device 12 as the relative orientation of magnetic moments in layers 18 and 22.
A conductive line 24 is shown over layer 22, and such conductive line extends into and out of the plane of the page. Conductive line 24 can comprise, for example, one or more metals and/or metal alloys, including, for example, copper and/or aluminum.
An insulative material 26 extends over conductive line 16, and along the sides of bit 12 and conductive line 24. Insulative material 26 can comprise, for example, BPSG.
The construction 10 is an exemplary MRAM construction, and it is to be understood that various modifications can be made to the construction 10 for various applications. For instance, one or more electrically insulative layers (not shown) can be provided between device 12 and one or both of conductive lines 16 and 24. Also, one or more magnetic layers (not shown) can be stacked within device 12 in addition to the shown layers 18 and 22.
In operation, data is written to MRAM device 12 by passing current along the conductive lines 16 and 24 to change the relative magnetic orientation of layers 18 and 22 (i.e., to flip the relative orientation from parallel to anti-parallel, or vice versa). In theory, the relative orientation of layers 18 and 22 can be flipped by passing sufficient current along only one of lines 16 and 24, but in practice it is generally found to be advantageous to utilize both of lines 16 and 24 in writing information to device 12. Specifically, some current is initially passed along one of the lines 16 and 24 to induce a magnetic field in device 12 which starts to flip the relative magnetic orientation of layers 18 and 22, and then current is passed along the other of layers 16 and 24 to complete the flip of the relative magnetic orientation within device 12.
The operation of reading information from device 12 can utilize either inductive sensing or magnetoresistive sensing to detect the relative magnetic orientation of layers 18 and 22 within the device. The reading can utilize one or both of lines 16 and 24, and/or can utilize a separate conductive line (not shown).
It is advantageous to have lines 16 and 24 be orthogonal to one another at the location of device 12 to maximize the complementary effect of utilizing both of conductive lines 16 and 24. A device which utilizes a pair of independently controlled conductive lines for writing to and/or reading from an MRAM device is typically referred to as a half-select MRAM construction.
As discussed above, a single MRAM device can store a single bit of information. Accordingly, in applications in which it is desired to process multiple bits of information it is generally desired to utilize a plurality of MRAM devices, with each of the devices independently storing bits of information. The devices will typically be arranged in an array, and an exemplary array 50 of MRAM devices is illustrated in FIG. 2. The array comprises individual MRAM devices which are shown schematically as 52, 54, 56, 58, 60, 62, 64, 66, 68, 70, 72 and 74. Exemplary magnetic poles associated with layers of the MRAM devices (such as the poles associated with layers 18 or 22 of FIG. 1) are illustrated with a “+” and “−” to indicate directional orientations of the poles.
A problem which can occur during operation of the array 50 of MRAM devices is that cross-talk can occur between adjacent devices such that a magnetic field of one device influences the magnetic fields of one or more neighboring devices. The cross-talk can disrupt reading and writing operations to individual MRAM devices, and, in particularly problematic instances, can change a value of a stored bit within an MRAM device. Accordingly, it is desired to alleviate, and preferably prevent, cross-talk between neighboring MRAM devices of a memory array.